Hi Martinj: thanks for looking into this! I think have a better fix for these:
the problem is actually in the AssertingAtomicReaders that AssertingDirectoryReader wraps its subreaders with. So I added the invisible-cache-key hack there, and removed it completely from LuceneTestCase. I tested this with the hudson seeds that failed (at their appropriate revisions) and it seems to work fine. I also ran tests for queries/grouping/join with -Dnightly=true, -Dtests.multiplier=5, etc etc a few times and it all works. I'd really like to have AssertingDirectoryReader being used again. If there are problems we can just back out the change. On Thu, Jul 19, 2012 at 5:48 AM, <[email protected]> wrote: > Author: mvg > Date: Thu Jul 19 09:48:04 2012 > New Revision: 1363272 > > URL: http://svn.apache.org/viewvc?rev=1363272&view=rev > Log: > Fix of rare FC insanity during tests that have occurred in grouping & joining > tests. > > Modified: > > lucene/dev/trunk/lucene/test-framework/src/java/org/apache/lucene/util/LuceneTestCase.java > > Modified: > lucene/dev/trunk/lucene/test-framework/src/java/org/apache/lucene/util/LuceneTestCase.java > URL: > http://svn.apache.org/viewvc/lucene/dev/trunk/lucene/test-framework/src/java/org/apache/lucene/util/LuceneTestCase.java?rev=1363272&r1=1363271&r2=1363272&view=diff > ============================================================================== > --- > lucene/dev/trunk/lucene/test-framework/src/java/org/apache/lucene/util/LuceneTestCase.java > (original) > +++ > lucene/dev/trunk/lucene/test-framework/src/java/org/apache/lucene/util/LuceneTestCase.java > Thu Jul 19 09:48:04 2012 > @@ -1048,7 +1048,7 @@ public abstract class LuceneTestCase ext > if (r instanceof AtomicReader) { > r = new FCInvisibleMultiReader(new > AssertingAtomicReader((AtomicReader)r)); > } else if (r instanceof DirectoryReader) { > - r = new FCInvisibleMultiReader(new > AssertingDirectoryReader((DirectoryReader)r)); > + r = new FCInvisibleMultiReader((DirectoryReader)r); > } > break; > default: > > -- lucidimagination.com --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
